{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ARXVBVSCT5SK5NKPMRH2CDTPY2","short_pith_number":"pith:ARXVBVSC","schema_version":"1.0","canonical_sha256":"046f50d6429f64aeb54f644fa10e6fc6aa75abc72dc1288a6ba73ff892faa06a","source":{"kind":"arxiv","id":"2605.21994","version":1},"attestation_state":"computed","paper":{"title":"Ex-GraphRAG: Interpretable Evidence Routing for Graph-Augmented LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arvindh Arun, Maya Bechler-Speicher, Rishi Puri, Steffen Staab, Yoav Kor Sade","submitted_at":"2026-05-21T04:43:08Z","abstract_excerpt":"GraphRAG conditions language models on subgraphs retrieved from knowledge graphs, encoded via message-passing GNNs. Because these encoders entangle node contributions through iterated neighborhood aggregation, there is no closed-form way to determine how much each retrieved entity influenced the encoder's output, and therefore no way to faithfully audit what structural evidence actually reached the model. We introduce Ex-GraphRAG, which replaces the GNN encoder with a Multivariate Graph Neural Additive Network (M-GNAN), an extension of additive graph models to high-dimensional embedding spaces"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.21994","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-21T04:43:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e443b54c29d928fc1e5af486a99dbf7124403d87451d1216f526375f0d104ca3","abstract_canon_sha256":"d23c0b63db19502cc36d4bb64abe7e86ce642edf83987a5c6703a04c62e7a3a5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:04:19.532522Z","signature_b64":"aSyflbqa/E2F6iSbwHrxGoqky0J6JQ/VVZrwyvsEya7tyRWeNBMVS6GWybh6nmVdr4wxVFnNT8H2Fw3DvY0EAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"046f50d6429f64aeb54f644fa10e6fc6aa75abc72dc1288a6ba73ff892faa06a","last_reissued_at":"2026-05-22T01:04:19.531846Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:04:19.531846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Ex-GraphRAG: Interpretable Evidence Routing for Graph-Augmented LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arvindh Arun, Maya Bechler-Speicher, Rishi Puri, Steffen Staab, Yoav Kor Sade","submitted_at":"2026-05-21T04:43:08Z","abstract_excerpt":"GraphRAG conditions language models on subgraphs retrieved from knowledge graphs, encoded via message-passing GNNs. Because these encoders entangle node contributions through iterated neighborhood aggregation, there is no closed-form way to determine how much each retrieved entity influenced the encoder's output, and therefore no way to faithfully audit what structural evidence actually reached the model. We introduce Ex-GraphRAG, which replaces the GNN encoder with a Multivariate Graph Neural Additive Network (M-GNAN), an extension of additive graph models to high-dimensional embedding spaces"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.21994","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.21994/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.21994","created_at":"2026-05-22T01:04:19.531970+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.21994v1","created_at":"2026-05-22T01:04:19.531970+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.21994","created_at":"2026-05-22T01:04:19.531970+00:00"},{"alias_kind":"pith_short_12","alias_value":"ARXVBVSCT5SK","created_at":"2026-05-22T01:04:19.531970+00:00"},{"alias_kind":"pith_short_16","alias_value":"ARXVBVSCT5SK5NKP","created_at":"2026-05-22T01:04:19.531970+00:00"},{"alias_kind":"pith_short_8","alias_value":"ARXVBVSC","created_at":"2026-05-22T01:04:19.531970+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2","json":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2.json","graph_json":"https://pith.science/api/pith-number/ARXVBVSCT5SK5NKPMRH2CDTPY2/graph.json","events_json":"https://pith.science/api/pith-number/ARXVBVSCT5SK5NKPMRH2CDTPY2/events.json","paper":"https://pith.science/paper/ARXVBVSC"},"agent_actions":{"view_html":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2","download_json":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2.json","view_paper":"https://pith.science/paper/ARXVBVSC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.21994&json=true","fetch_graph":"https://pith.science/api/pith-number/ARXVBVSCT5SK5NKPMRH2CDTPY2/graph.json","fetch_events":"https://pith.science/api/pith-number/ARXVBVSCT5SK5NKPMRH2CDTPY2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2/action/storage_attestation","attest_author":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2/action/author_attestation","sign_citation":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2/action/citation_signature","submit_replication":"https://pith.science/pith/ARXVBVSCT5SK5NKPMRH2CDTPY2/action/replication_record"}},"created_at":"2026-05-22T01:04:19.531970+00:00","updated_at":"2026-05-22T01:04:19.531970+00:00"}